Homesource Forums

When hyperspaceing the effects that "flicker" at the sides of the hyperspace window stretch across the screen.
Ion cannon beams don't render most of the time and when they do they are in the wrong location
The field that gets emitted by harvesters and healing ships doesn't render most or the time and if it does in the wrong location.

Image:
http://img5.imageshack.us/img5/8702/hyperspacebug.png

I tried to log this issue with Manits but when I submitted it I got the error:


As far as I could see there is no Affected field and pressing back didn't highlight an issue (which Mantis said it would do)

I think Ive seen the ion cannon problem before, If what i saw was what you are describing. I had a block of code in the middle of etgEffectDraw that told it to return prematurely. I dont have the block of code anymore, (for obvious reasons), but i believe it did something like this.

Psudo Code


I put the block right before the call of the partRenderSystem() function. (about line 2396)

the address that I was testing for was one that commonly popped up when ion cannon beams fired. the address would eventually find itself as the handle in the function trMakeCurrent and cause a crash....(it would come up as a null pointer).

hope this helps.

I take it back... I found the block. it still resides in render.c under the rndMainViewRenderFunction() i just commented it out.

it looks like this.


the addresses I was checking ultimately ended up in trMakeCurrent() as invalid handles, and i was trying to figure out how far back the problem originates... I believe my problem was actually in etgEffectDraw someplace (I havent found it yet)...

the function trace looks like this.
rndMainViewRenderFunction() //render.c
etgEffectDraw() //ETG.c
partRenderSystem() //Particle.c
partRenderMeshSystem() //Particle.c
partMeshMaterialPrepare() //Particle.c
trMakeCurrent() //texreg.c

returning at rndMainViewRenderFunction instead of trMakeCurrent caused an ion cannon bug similar to the one you are describing above. so I would imagine one of these functions is to blame.

edit: also my logic was flawed when i wrote this block. effect->collMyBlob does not become the handle down the line, it just has the same address at this point.

Still no luck adding this issue to mantis, but I did add something "new" to the wishlist . Feel free to assign it to me!

I've verified the problem on Linux, seems that it was introduced in rev 659.

basicmark: you want to go ahead and write it up in mantis?

Thanks zapkitty, mantis accepts the bug report now. I also updated my feature request

I used the "move issues" button to move this bug and the ARM issue from "wishlist" to "Homeworld SDL"...

... I'd feel so smart if it hadn't taken me so long to figure it out

Oh.

r659. That was me then.

Aunxx.

Possibly a compiler issue.

Please see
http://homesource.nekomimicon.net/sourceforum/viewtopic.php?f=8&t=313
for more info.

Aunxx